LAUSR

Abstract In this paper we propose a zero-inefficiency stochastic frontier model with a simple semiparametric approach using panel data. We model the frontier with a smooth coefficient function and specify a nonzero conditional probability for firms being fully efficient to be a known function of environment variables. Following (Yao et al., 2017) we propose a three step semiparametric estimator which is computationally efficient. The simulation results reveal encouraging finite sample properties. We illustrate the applicability of our model using country level data from the Penn World Table.